Factors affecting foreign exchange rate?

Introduction

In today’s world, domestic economic growth depends not only on local productivity and wealth, but also heavily on international trading. However, international trading is much more complicated than domestic trading because of many additional factors such as exchange rate, tax, tariffs, trade embargo, etc… In this project, I focus on the analysis on the exchange rate because it is the first risk source that foreign trading faces and currencies are usually very liquid in the foreign exchange market, or the FX market. In fact, the FX market is the largest market in the world. In 2001, it is estimated that $1210 billion was traded per day, that is, roughly $200 dollars per person in the whole world.

I chose to study the FX rate between Japan and US for several reasons. Firstly, these two currencies are the 2nd and 3rd currencies that dominates the FX market, with 17% in $USD and 10% in Japanese Yen. This suggests that the currency trading between these two are highly volatile and thus efficient because of the enormous amount of trading taking place simultaneously. Secondly, both countries are well developed countries with healthy economies and experienced traders, so it is reasonable to compare the two. Lastly, I wanted to investigate the relationship among the factors affecting the FX between two countries with an economical dependent structure: Japanese market is highly dependent on the export to US but in the opposite direction, the dependency is not that high. It is of interest to examine whether this economical polarization has any impact on the relationships among those factors on FX.

This project is divided into two parts: the first part studies the ingredients affecting the FX. In the beginning, I considered a large set of possible factors, including global factors that affect both countries and local factors that affect only the individual country. Then I reduced the set by performing a regression on the FX over these factors and try to see if I was able to obtain a smaller, yet meaningful subset of these factors that can explain the movement of the FX well. The second part focuses on the relationships among those significant factors and how the market data draws a division of them into smaller subgroup for us to investigate deeper properties. I used principle component analysis that alludes this division and then calculated the partial correlation as suggested by the division. Finally, using discriminant analysis, I looked for various economic aspects that indicate economic growth of a country.


Part I

In traditional financial engineering literature, the FX model depends only on the difference between the domestic and foreign short rate. In particular, the simplest, yet widely used model, is the following:

Here F denotes the value of one unit of foreign currency in terms of one unit of domestic currency, rd and rf denotes the domestic and foreign short rate respectively, and finally F and Zt is the volatility and the random Brownian process that governs the dynamics of the exchange rate. It is easy to see that this simple model only utilize the short rate difference, while the other possible factors are hidden in the random process.